Nuclear-L1 norm joint regression for face reconstruction and recognition with mixed noise

作者：

Highlights：

• We use matrix distributions to describe the mixed noise and structural noise.

• We derive two nuclear-L1 norm joint matrix regression models.

• We solve the proposed models by using ADMM, and provide the complexity analysis.

• Experimental results demonstrate the advantages of our methods.

摘要

•We use matrix distributions to describe the mixed noise and structural noise.•We derive two nuclear-L1 norm joint matrix regression models.•We solve the proposed models by using ADMM, and provide the complexity analysis.•Experimental results demonstrate the advantages of our methods.

论文关键词：Sparse representation,Nuclear norm,Dependence,Alternating direction method of multipliers (ADMM),Face recognition

论文评审过程：Received 12 February 2015, Revised 6 May 2015, Accepted 23 June 2015, Available online 2 July 2015, Version of Record 19 August 2015.

论文官网地址：https://doi.org/10.1016/j.patcog.2015.06.012